mlops-monitoring-drift

MLOps drift monitoring workflow for detecting data drift, concept drift, and quality degradation with actionable response rules. Use when production ML systems need drift detection thresholds and escalation ownership; do not use for model-architecture research decisions.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "mlops-monitoring-drift" with this command: npx skills add kentoshimizu/sw-agent-skills/kentoshimizu-sw-agent-skills-mlops-monitoring-drift

Mlops Monitoring Drift

Overview

Use this skill to detect meaningful model degradation early and trigger appropriate remediation actions.

Scope Boundaries

  • Use this skill when the task matches the trigger condition described in description.
  • Do not use this skill when the primary task falls outside this skill's domain.

Shared References

  • Drift alerting and escalation rules:
    • references/drift-alerting-escalation-rules.md

Templates And Assets

  • Drift monitoring policy template:
    • assets/drift-monitoring-policy-template.md

Inputs To Gather

  • Drift signals and quality metrics to monitor.
  • Alert thresholds and acceptable noise level.
  • Escalation owners and response SLA.
  • Retraining and rollback policies.

Deliverables

  • Drift monitoring policy and thresholds.
  • Alert routing and severity model.
  • Response playbook for drift events.

Workflow

  1. Define monitoring policy in assets/drift-monitoring-policy-template.md.
  2. Validate threshold actionability via references/drift-alerting-escalation-rules.md.
  3. Test alert behavior with historical replay or backtests.
  4. Assign response ownership and SLA per severity.
  5. Publish retraining/mitigation decision criteria.

Quality Standard

  • Alerts are actionable, not noise-heavy.
  • Severity levels map to clear response ownership.
  • Retraining triggers are explicit and auditable.

Failure Conditions

  • Stop when drift thresholds are not operationally actionable.
  • Stop when alerts have no clear owner.
  • Escalate when degradation risk remains unmanaged.

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Research

risk-requirements-analysis

No summary provided by upstream source.

Repository SourceNeeds Review
Research

ux-research-synthesis

No summary provided by upstream source.

Repository SourceNeeds Review
Research

architecture-tradeoff-analysis

No summary provided by upstream source.

Repository SourceNeeds Review
Research

algorithm-complexity-analysis

No summary provided by upstream source.

Repository SourceNeeds Review